Parameter reference

A MILEVA run is configured by three mutable structs that main_SphCon receives together:

Struct	File	Holds
SimConfig	config/SimConfig.jl	Geometry, resolution, solvers, physics switches, time stepping, output. This page documents every field.
MaterialProperties	config/MaterialProperties.jl	Layer radii, creep-law constants, yield parameters, phase-transition slopes and density jumps, composition viscosity factors. Summarised at the end of this page.
TemperatureConditions	config/TemperatureConditions.jl	Boundary temperatures, initial profile, CMB patches. Summarised at the end of this page.

All three are @kwdef structs, so a configuration is plain Julia: build the defaults, overwrite fields, then call validate! on each before the run.

using Plots   # visualization/plot_panels.jl needs Plots loaded first
include("include/mileva.jl"); include("main_SphCon.jl")

cfg  = SimConfig(Nr = 64, Nt = 512)        # keywords at construction
cfg.geometry   = "spherical_annulus"       # or assignment afterwards
cfg.solverType = "mumps"
mat  = MaterialProperties()
temp = TemperatureConditions()
validate!(cfg); validate!(mat); validate!(temp)

history = main_SphCon(cfg, mat, temp)

A parameter file for run_mileva.jl or bin/run_mpi.jl is a Julia script that assigns to cfg, mat and temp in this way.

How to read the tables

The Default column is the value in the @kwdef declaration. Read by names the files that read the field (found by searching the source, excluding tests/ and benchmarks/). The status column flags anything that does not do what its name suggests:

• live: read by the time loop or a solver it calls.

• unused: no code outside config/SimConfig.jl reads it.

• refused: validate!(cfg) raises an error if it is set, because the feature behind it is not implemented on the time loop.

• inert: read only by code that no entry point includes (the AMR prototype under solvers/amr/), or only by a function nothing calls.

• helper only: read by a helper function that the standard time loop does not call; a custom driver can call it.

• experimental: live but not recommended; the trade-off is described.

Set what a run depends on

The struct defaults describe an Earth-sized, ALA, composite-capable shell (R_in = 3481 km, R_out = 6371 km, compressibilityMode = "ALA", solverType = "direct", Nr × Nt = 192 × 512). A benchmark sets every switch it depends on explicitly, as benchmarks/hernlund_tackley_annulus.jl does (it clears all optional flags in a loop, then sets the ones it needs).

Dimension, geometry and grid

The 2-D shell as the defaults define it: R_in and R_out from SimConfig, T_cmb and T_surf from TemperatureConditions. The blue sector shows cell faces built by setup_grid on a coarse grid (Nr = 12, Nt = 96) with amrStretchFactor = 2, which narrows the cells toward both walls; the shading from hot to cold is illustrative only.

Field	Default	Units	Status	Meaning
dim	2		live	2: spherical (or cylindrical) annulus. 3: Yin-Yang spherical shell.
geometry	"spherical_annulus"		live	2-D radial metric. "spherical_annulus" uses the measure $r^{2}dr$; "cylindrical_annulus" uses $rdr$. "Cartesian" is reserved and unused by the annulus path. validate! normalises "sphere"/"spherical" and "cylinder"/"cylindrical" and refuses the bare "annulus" as ambiguous. Read through geometry_exponent(cfg) in utils/geometry.jl.
R_in	3481e3	m	live	Inner radius (core-mantle boundary).
R_out	6371e3	m	live	Outer radius (surface).
Nr	192		live	Radial cells.
Nt	512		live	2-D: azimuthal cells over the full $2\pi$. 3-D: colatitude cells Nth of each Yin-Yang patch; the longitude count is derived as Nph = 3·Nth·(chunkPhiRange/(3π/2)), rounded to a multiple of 3 (_setup_grid_3d in utils/setup_grid.jl).
chunkPhiRange	3π/2	rad	live (3-D)	Longitude extent of the Yin patch. Below $3\pi /2$ the domain is a chunk and the Yin-Yang Schwarz coupling must be off.
amrStretchFactor	0.0		live	Static tanh stretching of the radial grid toward both boundaries, $\beta$ in _amr_radial_grid (utils/setup_grid.jl). 0 is uniform. Works in 2-D and 3-D. Despite the name this is not adaptive refinement.
amrStretchTheta	0.0		live (3-D)	Same tanh stretching applied to colatitude.
amrStretchPhi	0.0		live (3-D)	Same tanh stretching applied to longitude.

The geometry exponent $m$ ($m=1$ cylindrical, $m=2$ spherical) is the only thing that differs between the two 2-D geometries; the mesh is identical. It enters the radial divergence, the cell measure, the boundary areas and the conductive Nusselt baseline:

$${\mathrm{\nabla }}_r\cdot f=\frac{1}{r^m}\frac{\partial (r^{m}f)}{\partial r},|\mathrm{c}\mathrm{e}\mathrm{l}\mathrm{l}|=\mathrm{\Delta }\theta {\int }_{r_{i-1/2}}^{r_{i+1/2}}{r}^{m}dr.$$

For a Venus-sized shell the inner-to-outer boundary area ratio is 0.514 cylindrical against 0.264 spherical (geometry_area_ratio), which is why the cylindrical setting is not used for science runs.

Physical constants

Field	Default	Units	Status	Meaning
g	9.81	m s⁻²	live	Gravitational acceleration (uniform).
alpha	3e-5	K⁻¹	live	Reference thermal expansivity (Boussinesq buoyancy).
rho0	3300.0	kg m⁻³	live	Reference density.
kappa	1e-6	m² s⁻¹	live	Reference thermal diffusivity. Under ALA and TALA the diffusivity grows with depth as $\kappa (1+f{z}^2)$, $f$ = kappaDepthFactor (thermal_diffusivity in physics/compressibility.jl).
kappaDepthFactor	0.5		live	Depth enhancement $f$ of the diffusivity under ALA and TALA, with $z$ the normalised depth. 0 keeps $\kappa$ constant, as the compressible example does to compare the formulations with Boussinesq.
SEC_PER_MYR	3.1557e13	s	derived	Overwritten to 3.15576e13 by validate!.
R_GAS	8.314	J mol⁻¹ K⁻¹	derived	Gas constant; reset by validate!.

Compressibility and the energy equation

Field	Default	Units	Status	Meaning
compressibilityMode	"ALA"		live	"Boussinesq", "ALA" (anelastic liquid) or "TALA" (truncated ALA, no adiabatic heating). Benchmarks set "Boussinesq".
useCompressibility	true		unused	Legacy alias. compressibilityMode is the switch that is read.
T_potential_ref	1350.0	K	live	Potential temperature of the ALA reference adiabat; also used by the marker initial profile.
adiabat_gradient	0.3e-3	K m⁻¹	unused	No reader. The adiabat is integrated from alpha_surf_ref, alpha_cmb_ref and gruneisen.
alpha_surf_ref	3e-5	K⁻¹	live	Reference expansivity at the surface (ALA profile).
alpha_cmb_ref	1e-5	K⁻¹	live	Reference expansivity at the CMB (ALA profile).
gruneisen	1.1		live	Grüneisen parameter of the reference state; read only under ALA/TALA.
referenceTable	""		live	Perple_X T-P table whose adiabat from T_potential_ref gives the ALA/TALA reference state ($T_{\text{ad}}$, ${\rho }_{\text{ref}}$, $\alpha$); see Material tables.
useShearHeating	true		live	Carries viscous dissipation and the adiabatic term of the energy equation together under ALA. The two O(Di) terms cancel in the volume integral of a steady layer, so including one alone injects or removes that whole power.
useCompressibleStokes	false		live (3-D)	3-D continuity becomes $\mathrm{\nabla }\cdot ({\rho }_{\mathrm{r}\mathrm{e}\mathrm{f}}\mathbf{v})=0$. Breaks the $G=-B^{\mathsf{T}}$ symmetry of the incompressible assembly.

Temperature transport

Field	Default	Units	Status	Meaning
temperatureOnGrid	false		live (2-D)	false: temperature rides on the markers; the grid solves diffusion and hands back the increment. true: temperature is a grid field advected by advect_temperature_2d (utils/advect_temperature.jl).
subgridDiffusivity	1.0		live (2-D)	Gerya & Yuen (2003) sub-grid relaxation coefficient $d$ (subgrid_temperature_correction!, markers/subgrid_diffusion.jl). Required for the marker-temperature path; 0 disables.
tempAdvectionScheme	"upwind"		live	"upwind" (first order) or "TVD" (van Leer flux-limited) for grid advection in 2-D when temperatureOnGrid = true. First-order advection is very diffusive: set "TVD" whenever temperature is on the grid. The 3-D Yang patch temperature is advected with advect_temperature_3d_upwind! regardless of this field.
useImplicitTemp	true		live	Crank-Nicolson/theta diffusion (false: explicit Euler).
implicitTheta	0.5		live	Theta weight: 0.5 Crank-Nicolson, 1.0 backward Euler.
deliveryScaleMax	2.0		live	Cap on the conservative marker hand-back factor (cell measure / marker measure, delivery_scale in markers/interp_grid_to_markers.jl). Uncapped, a wall cell with few markers over-delivers the diffusion increment and can blow the run up. Energy the cap withholds appears in budget_resid of diagnostics.csv.

The temperature boundary conditions are Dirichlet values held at the wall faces through a half-cell ghost in diffuse_temperature (solvers/diffuse_temperature.jl); no flow cell is overwritten.

Stokes operator

Field	Default	Units	Status	Meaning
useDeviatoricStress	true		live (2-D)	Assembles $\mathrm{\nabla }\cdot (2\eta \dot{\mathit{\epsilon }})$ as the factored product of a constitutive map and a discrete divergence (solvers/stokes_assembly.jl). false selects $\eta {\mathrm{\nabla }}^2\mathbf{v}$, which equals it only at constant viscosity.
useCurvatureTerms	true		live (3-D)	Selects the full spherical deviatoric stress in assemble_stokes_3d. false selects the componentwise scalar Laplacian.
useGeometricEta	true		live (3-D)	Geometric rather than arithmetic interpolation of viscosity to faces and edges.
velReg	1e-9		live	Diagonal regularisation of the momentum rows (dimensionless). The operator tests set it to 0.
pPen	0.0		live	Pressure penalty on the continuity rows ($\mathrm{\nabla }\cdot \mathbf{v}=-p_{\mathrm{P}\mathrm{e}\mathrm{n}}P/(r\mathrm{\Delta }\theta )$). Keep 0: the staggered grid is inf-sup stable, and a penalty of 1e-6 left a measurable divergence in every cell.
removeNetRotation	true		live (2-D)	After each solve, subtract the rigid rotation $\omega r$ with $\omega =\int \rho r{V}_{t}dV/\int \rho r^{2}dV$, $dV=r^{m}drd\theta$ ($\rho ={\rho }_{\text{ref}}$ under ALA/TALA, 1 under Boussinesq), so the angular momentum vanishes. Has no effect in 3-D; validate! warns once if it is left on there.
Vmax_limit	1.6e-8	m s⁻¹	live	Threshold for the [VELOCITY WARNING] diagnostic (about 50 cm/yr). Also the hard cap when useVelocityLimit = true.
useVelocityLimit	false		live	Legacy hard cap that rescales the whole $(\mathbf{v},P)$ field. It invalidates the dynamics; leave it off.
picardMax	0		live	Picard sweeps per step for the nonlinear viscosity. 0 = auto: 50 when a yield law is active, 1 otherwise. "Active" means any of mat.maxYieldStress, mat.cohesion_upper, mat.cohesion_lower is finite, which holds for the default material even on the simple viscosity path; set picardMax = 1 when the viscosity does not depend on strain rate.
picardTol	1e-4		live	Relative change in $V_{max}$ that ends the Picard loop.
picardDamping	1.0		live	Log-space blending weight between sweeps; 1 is undamped.

Stokes solver

Field	Default	Units	Status	Meaning
solverType	"direct"		live	2-D: "direct" (UMFPACK, solve_stokes_v2), "mumps" (solve_stokes_mumps, recommended for 2-D), "amg" (solve_stokes_amg), "multigrid" (2-D V-cycle). 3-D: the time loop always calls multigrid_stokes; set "multigrid" explicitly.
mgLevels	4		live	Multigrid levels. In 3-D use mgLevels ≥ 2. With one level multigrid_stokes falls back to a direct LU of the full 3-D operator, whose memory grows past a 16 GB machine at modest resolution.
mgVcycles	2		live	V-cycles per application (2-D multigrid).
mgSmooth	2		live	Pre/post SCGS smoothing sweeps.
krylovTol	1e-8		live	Relative tolerance of the iterative solvers (FGMRES/GMRES).
krylovMaxIter	200		live	Maximum Krylov iterations.
luCacheTol	1e-4		live (2-D, MUMPS)	Reuse the MUMPS factor while the largest relative change of the viscosity stays below this.
luReuseMaxIter	0		live (2-D, MUMPS)	When the viscosity has moved, precondition GMRES with the stale factor for up to this many iterations before refactorising. 0 disables.
luReuseTol	1e-11		live (2-D, MUMPS)	GMRES tolerance of the stale-factor solve.

Yin-Yang coupling (3-D)

Field	Default	Units	Status	Meaning
useFullYinYangCoupling	true		live	Couple both the θ-walls and the φ-walls of each patch in the Schwarz iteration (yy_compute_th_bcs, yy_compute_ph_bcs in utils/yy_interp.jl). false couples θ-walls only.
nTSchwarz	3		live	Temperature Schwarz sweeps per step between Yin and Yang. Read through getfield(cfg, :nTSchwarz) in main_SphCon.jl.
useBicubicYinYang	false		experimental	Catmull-Rom bicubic sampling of the peer patch. In development tests it gave a larger equatorial seam than bilinear at Nt = 12.
δ_yy	0	cells	experimental	Width of an overlap buffer added outside each patch's natural domain. 0 is the production setting.
useMinOverlapGrid	false		experimental	Minimum-overlap ("baseball border") Yin-Yang grid with redundant cells eliminated. Reduces the equatorial seam but stalls FGMRES at low viscosity contrast. Not a true ragged grid; see the notes in utils/yy_interp.jl.

MPI decomposition (3-D)

Field	Default	Units	Status	Meaning
useMPI	false		live (3-D driver) / refused in 2-D	Selects 3-D domain decomposition through bin/run_mpi.jl and solve_stokes_distributed!. validate! refuses useMPI = true with dim = 2: there is no 2-D decomposition, and the 2-D MUMPS path initialises MPI by itself.
mpiNr, mpiNt, mpiNp	1, 1, 1		live	Ranks per Yin-Yang patch in r, θ, φ. Total ranks = 2 × mpiNr × mpiNt × mpiNp (mpi_init! in parallel/mpi_init.jl).

Viscosity

Bounds and simple laws

Field	Default	Units	Status	Meaning
eta_min	1e18	Pa s	live	Lower viscosity clamp.
eta_max	1e24	Pa s	live	Upper viscosity clamp.
useFrankKamenetskii	false		live	Frank-Kamenetskii viscosity (frank_kamenetskii_viscosity!, utils/fk_viscosity.jl).
fkForm	"linearized"		live	"linearized": $\eta ={\eta }_{\mathrm{r}\mathrm{e}\mathrm{f}}\exp [E/(1+T/T_{\mathrm{r}\mathrm{e}\mathrm{f}})-E/2]$. "simple": $\eta ={\eta }_{\mathrm{r}\mathrm{e}\mathrm{f}}\exp [-{\theta }_{FK}(T-T_{\mathrm{t}\mathrm{o}\mathrm{p}})/\mathrm{\Delta }T]$, contrast $e^{{\theta }_{FK}}$ across the layer.
fk_E	29.95		live	Nondimensional activation energy (linearized form).
fk_T_ref	2500.0	K	live	Reference temperature (linearized form).
fk_theta	6.9		live	${\theta }_{FK}$ of the simple form. The H&T variable-viscosity cases use $\ln {10}^6=13.8155$.
fk_eta_ref	4.5e19	Pa s	live	Reference viscosity (simple form: the value at the top temperature).
useArrhenius	false		helper only	$\eta ={\eta }_{\mathrm{r}\mathrm{e}\mathrm{f}}\exp [(E/R)(1/T-1/T_{\mathrm{r}\mathrm{e}\mathrm{f}})]$, applied by arrhenius_viscosity! (utils/arrhenius_viscosity.jl, 3-D arrays). main_SphCon does not call it, so in the standard time loop this flag and the three arrhenius_* fields change nothing; a driver has to call the helper itself.
arrhenius_E	300e3	J mol⁻¹	helper only	Activation energy.
arrhenius_T_ref	1600.0	K	helper only	Temperature at which $\eta ={\eta }_{\mathrm{r}\mathrm{e}\mathrm{f}}$.
arrhenius_eta_ref	1e21	Pa s	helper only	Reference viscosity.
use660kmJump	false		live (composite)	Lower-mantle viscosity offset by eta660kmFactor below depth660km. Live inside composite_rheology, which evaluates it as a creep prefactor at the interface. On the simple path it needs the helper apply_660km_jump! (utils/viscosity_jumps.jl), which main_SphCon does not call.
eta660kmFactor	30.0		live	Lower-mantle viscosity factor.
depth660km	660e3	m	live	Depth of the jump below R_out.

In the time loop (main_SphCon.jl), cfg.useCompositeRheology selects composite_rheology; otherwise the viscosity is compute_viscosity (physics/compute_viscosity.jl), a layered law $\eta ={\eta }_0^{\mathrm{l}\mathrm{a}\mathrm{y}\mathrm{e}\mathrm{r}}\exp [E(1/T-1/T_{\mathrm{r}\mathrm{e}\mathrm{f}})]$ with $E$ in kelvin taken from mat.E_act_scale (plus mat.V_act_scale times lithostatic pressure in 2-D) and the *_eta0 layer values of MaterialProperties, after which frank_kamenetskii_viscosity! overwrites it when useFrankKamenetskii is set, and depletion stiffening is applied when useDepletionStiffening is set. mat.E_act_scale = 0 gives an isoviscous layer, which is how the Hernlund & Tackley benchmark pins the viscosity. | radialViscosityFile | "" | | live | Viscosity-against-depth profile; replaces the simple viscosity path (Material tables). | | lateralViscosityFile | "" | | live | $H/nR$ against depth for the lateral temperature factor of the profile viscosity. | | maxLateralViscosityVariation | 1e3 | | live | Cap on that lateral factor, either way. |

Composite rheology and plasticity

Field	Default	Units	Status	Meaning
useCompositeRheology	false		live	Diffusion plus dislocation creep with optional Peierls, GBS and Drucker-Prager-type yielding (composite_rheology, physics/composite_rheology.jl). Creep and yield constants come from MaterialProperties.
useDepletionStiffening	false		live	Apply the depletion stiffening mat.harzburgite_eta_factor on the simple viscosity path as well, without switching on the composite branch.
useSimpleYield	false		live	Cap the simple-path viscosity at ${\tau }_y/(2{\dot{\epsilon }}_{II})$ with the Drucker-Prager yield stress from MaterialProperties (simple_yield!, physics/compute_viscosity.jl), iterated within the step; exercised in 2-D only; refused with useCompositeRheology or a radial viscosity file.
usePeierlsCreep	false		live	Add Peierls (low-temperature plasticity) creep where $T<1100$ K; the law is mat.flowlaw_peierls (Flow laws).
useGBS	false		live	Add grain-boundary sliding creep; with a published mantle flow law selected it is the Hirth & Kohlstedt (2003) dry olivine law.
gbs_A, gbs_n, gbs_m	6.5e-16, 2.0, 2.0		live	GBS prefactor, stress and grain-size exponents.
useLayeredActivationEnergy	false		live	Switch $E$, $V$ from upper- to lower-mantle values at the interface.
useFullPlasticity	false		live	Strain-accumulation feedback for yield weakening.
strainWeakeningFactor	1.0		live	Yield-strength factor at full weakening; 1 = no weakening.
strainWeakeningE1, strainWeakeningE2	0.5, 1.5		live	Accumulated strain at onset and saturation of weakening.
yieldMinT	0.0	K	live	Yielding disabled below this temperature; 0 = yield law active everywhere.
depletionStiffensYield	false		live	Also scale the yield viscosity by the depletion factor (legacy; not Byerlee consistent).
viscosityCombination	"min"		live	"min": $\eta =min({\eta }_{\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}},{\eta }_Y)$. "harmonic": $1/\eta =1/{\eta }_{\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}}+1/{\eta }_Y$.
continentViscosityAveraging	"factor"		live	"factor": continent field thresholded at 0.5, ductile factor mat.continent_eta_factor, yield parameters mixed linearly. "harmonic": mantle and continent viscosities evaluated and bounded separately, $1/\eta =(1-c)/{\eta }_m+c/{\eta }_c$ on the continent fraction $c$.
smoothViscosity	true		live	log-space smoothing of the composite viscosity (step 6).
smoothStrainRate	true		live	2-D: the rheology reads the smoothed strain-rate invariant; false reads the raw one.
yieldPressure	"lithostatic"		live	2-D yield-law pressure: the reference profile, or "total": that profile plus the solved Stokes pressure (top-row mean removed, floored at 0). Creep always uses the profile.
initialStrainRate	0.0	1/s	live	strain rate of the first viscosity of a fresh run; 0 starts the lid at eta_max, a reference rate starts it on the yield branch.
useViscoelastoplastic	false		live	Maxwell visco-elasto-plastic correction (viscoelastoplastic, physics/viscoelastoplastic.jl).
useAntisqueeze	false		live	Strain-rate correction in cells under resolved compression (antisqueeze, physics/antisqueeze.jl). Triggers only where $\mathrm{\nabla }\cdot \mathbf{v}<-{10}^{-3}{\dot{\epsilon }}_{II}$.
evolveGrainSize	false		live	Marker grain-size evolution (evolve_grain_size!, physics/evolve_grain_size.jl).
useDamage	false		live	Grid damage field with accumulation and healing (evolve_damage!, apply_damage_to_viscosity!, physics/damage.jl).
damage_acc_rate	1e14		live	Accumulation rate per unit strain.
damage_heal_time	3.15e15	s	live	Healing time (about 100 Myr).
damage_strain_rate_threshold	1e-17	s⁻¹	live	Strain rate below which damage does not grow.
damage_max_weakening	0.99		live	$\eta \to \eta (1-0.99D)$ at full damage.
useDruckerPrager	false		refused	utils/drucker_prager.jl is never called by the time loop; validate! raises an error. The yield law that runs lives in composite_rheology and takes its parameters from MaterialProperties (cohesion_upper, cohesion_lower, friction_upper, friction_lower, maxYieldStress).
dp_cohesion, dp_friction, dp_max_yield_stress, dp_damper_viscosity, dp_ref_strain_rate	55e6, 0.0, 55e6, 1e21, 1e-15	Pa, 1, Pa, Pa s, s⁻¹	inert	Read only by apply_drucker_prager!, which nothing calls.

Sticky air, free surface and isostasy

Field	Default	Units	Status	Meaning
useStickyAir	false		live	Override viscosity and density in a top layer (apply_sticky_air!, utils/sticky_air.jl).
airThickness	80e3	m	live	Sticky-air layer thickness.
airViscosity	1e18	Pa s	live	Sticky-air viscosity.
airDensity	1.0	kg m⁻³	live	Sticky-air density.
useFreeSurface	false		live	Kinematic surface tracking with theta-method stabilisation (Kaus et al. 2010) and a mesh-velocity correction to the top-boundary temperature advection (free_surface, physics/free_surface.jl).
freeSurfaceStabilization	0.5		live	Theta of the stabilisation (0 explicit, 1 implicit).
freeSurfaceMaxDrop	50e3	m	live	Maximum surface deflection.
useIsostasy	false		live	Approximate isostatic topography (isostasy, physics/isostasy.jl).
isostaticMode	"Airy"		live	"Airy" or "Flexural".
isostaticDepth_km	100.0	km	live	Compensation depth.
isostaticTau_yr	3000.0	yr	live	Relaxation time.
flexuralRigidity	0.0	N m	live	Flexural rigidity; 0 = compute from elasticThickness_km.
elasticThickness_km	30.0	km	live	Elastic thickness for the flexural mode.

Phase transitions

Field	Default	Units	Status	Meaning
useMeltingPhase	false		live	Call melt_phase (physics/melt_phase.jl) each step: melt fraction, phase index, latent heat.
usePostPerovskite	false		live	Include the post-perovskite transition.
ppvDepth_km	2700.0	km	live	Post-perovskite reference depth.
ppvClapeyron	10e6	Pa K⁻¹	live	Post-perovskite Clapeyron slope.
usePhaseLatentHeat	false		live	Latent heat from the material derivative of a smooth phase fraction, $\rho LD\mathrm{\Gamma }/Dt$. Changes results; start runs fresh.
useClaperonBuoyancy	false		live	Smooth Clapeyron buoyancy for the 410/660 analogues, using gamma_*, drho_*, clap_* from MaterialProperties (compute_buoyancy). The spelling of the field name is as in the code.
usePhaseTransitionDensity	true		live	Smooth (tanh) phase-function density jumps for the 520 km analogue, and for the 410/660 analogues when useClaperonBuoyancy is off, plus post-perovskite if usePostPerovskite. Active only when a phase field exists (useMeltingPhase = true); the integer phase index is not sampled (compute_buoyancy).
useEarthPhaseBoundaries	true		live	Assign phase indices at Earth transition pressures (melt_phase).
useEarthDensityProfile	true		unused	No reader.
compClapeyron_dP_410, compClapeyron_dP_660	0.0, 0.0	Pa	live	Shift of the transition pressure per unit basalt fraction. 0 disables.

Thermochemical piles

Field	Default	Units	Status	Meaning
useThermochemicalPiles	false		live	Seed and evolve basal pile material (thermochemical_piles!, physics/thermochemical_piles.jl). Off by default because it adds denser, triple-heated material.
tcPileInitTheta	[π-0.4, π+0.4, 2π-0.4, 2π+0.4]	rad	live	Initial pile angles, taken in pairs: edges of each pile in 2-D, (θ, φ) centres in 3-D.
tcPileHeightFrac	0.08		live	Initial pile height as a fraction of the mantle depth.
tcPileDensityAnomaly	150.0	kg m⁻³	unused	Mentioned in a comment only. The pile density excess actually used is mat.tcpile_drho.
tcPileViscMult	10.0		unused	No reader. The pile viscosity factor is mat.tcpile_eta_factor.

Composition, melting and crust

These fields act only when trackBasalt = true (the melting and basalt bookkeeping in basalt_evolution!, physics/basalt_evolution.jl) unless noted.

Field	Default	Units	Status	Meaning
trackBasalt	false		live	Melt production, extraction, eruption and intrusion, basalt and depletion on markers, melt ledger.
initial_basalt_fraction	0.0		live	Basalt fraction of every marker at start (0.2 gives a pyrolite-like start).
eclogiteTransitionDepth	60e3	m	live	Depth below which basalt carries the eclogite density excess.
basaltDensityAnomaly	200.0	kg m⁻³	live	Eclogite density excess.
useEclogiteKinetics	false		live	Rising eclogite reverts only when shallower than the transition and warmer than eclogiteBackT_K.
eclogiteBackT_K	1000.0	K	live	Back-transformation temperature.
eclogiteBackWidth_K	100.0	K	live	Smoothing width of that threshold.
crustDensityAnomaly	0.0	kg m⁻³	live	Crust lighter than the mantle reference by this amount (positive = lighter). 0 = off.
crustViscosityFactor	1.0		live	Crust viscosity multiplier on the simple viscosity path. 1 = off.
fertileFraction	0.30		live	Maximum extractable melt fraction; depletion is capped here.
depletionDensityAnomaly	50.0	kg m⁻³	live	Density deficit of fully depleted residue, applied as $\mathrm{\Delta }\rho min(C_{\mathrm{d}\mathrm{e}\mathrm{p}}/F_{max},1)$.
depletionSolidusRise	150.0	K	live	Solidus rise of fully depleted residue, scaled the same way.
useMagmaOceanFreezing	true		live	Legacy per-step surface-freezing basalt source. It is a per-step quantum, not a rate; set false unless reproducing old reference runs.
eruptionEfficiency	0.20		live	Fraction of extracted melt erupted; the rest is intruded.
intrusionDepth_km	60.0	km	live	Top of the depth band in which intruded melt is deposited (basalt_evolution!). Not read with twoPhaseFlow.
meltExtractionTime_yr	1.0e6	yr	live	Segregation timescale; extraction relaxes the melt fraction as $1-e^{-\mathrm{\Delta }t/\tau }$, independent of the step. 0 = instantaneous. Not read with twoPhaseFlow.
meltExtractMaxDepth_km	330.0	km	live	Melt is extracted only above this depth.
meltTemperatureIsPotential	false		live	In a Boussinesq run, compare the solidus with the real temperature on the adiabat through $T$.
meltSegregationMovesMarkers	false		live	Legacy: displace melting markers upward. Off.
useHeatPipe	true		live	Magmatic heat transport by eruption and intrusion.
latentHeatMelt	4.0e5	J kg⁻¹	live	Latent heat of melting.
radioHeatPartitionD	0.0		live	Partition coefficient of heat-producing elements into melt; 0 = uniform heating.
pressureCMB_target_GPa	0.0	GPa	live	If positive, a linearly depth-increasing density is used so the CMB pressure matches this target in the melting laws (mantle_pressure, physics/melt_phase.jl).

Two-phase flow

Darcy flow of melt through a compacting matrix, 2-D spherical annulus only (physics/two_phase.jl, solvers/two_phase_assembly.jl; see Two-phase flow of melt and solid). Every field except the switch is read only when twoPhaseFlow = true; with the switch off a run is unchanged. validate! refuses the switch with 3-D, a non-spherical geometry, solverType other than "direct" or "mumps", compressibilityMode other than "Boussinesq", radioHeatPartitionD > 0, meltSegregationMovesMarkers, useStickyAir, useDeviatoricStress = false, useVelocityLimit = true, pPen other than 0, useAntisqueeze, and, with trackBasalt, useHeatPipe = false or useMagmaOceanFreezing = true.

Field	Default	Units	Status	Meaning
twoPhaseFlow	false		live	Solve for solid velocity, fluid pressure and compaction pressure with a porosity field; with trackBasalt, melt_two_phase! replaces basalt_evolution!.
meltDensityContrast	500.0	kg m⁻³	live	$\mathrm{\Delta }\rho ={\rho }_s-{\rho }_f$, melt buoyancy and Darcy driving density.
meltViscosity	1.0	Pa s	live	Melt viscosity $\mu$.
permeabilityK0	1e-9	m²	live	$k_0$ in $k=k_0{\varphi }^n$; sets the melt speed and so the cost.
permeabilityExponent	3.0		live	$n$ in $k=k_0{\varphi }^n$.
meltWeakeningAlpha	27.0		live	${\alpha }_{\varphi }$ in $\eta ={\eta }_s{e}^{-{\alpha }_{\varphi }\varphi }$.
bulkViscosityFactor	1.0		live	${\zeta }_0$ in $\zeta ={\zeta }_0\eta {\varphi }^{-m_{\zeta }}$.
bulkViscosityExponent	1.0		live	$m_{\zeta }$; with the defaults $\zeta =\eta /\varphi$.
twoPhasePorosityThreshold	1e-5		live	Cells at or below this porosity carry $p_c=0$ and are dry for the face mobility.
twoPhaseCFL	0.5		live	Courant number of the melt, Darcy-heat and compaction time-step limits.
initialPorosity	0.0		live	Uniform porosity at a fresh start.

Core-mantle boundary and core

Field	Default	Units	Status	Meaning
cmbHeatfluxMode	"fixed_T"		live	"fixed_T" (isothermal CMB), "fixed_flux", or "parameterized" (the core cools by the heat it delivers, cmb_heat_flux, physics/cmb_heat_flux.jl). Any other string is refused by validate!.
cmbHeatflux	0.0	W m⁻²	live	Flux for "fixed_flux".
coreDensity	12000.0	kg m⁻³	live	Mean core density (parameterized mode).
coreCp	800.0	J kg⁻¹ K⁻¹	live	Core heat capacity.
coreLatentFactor	1.0		live	Multiplier on the effective core heat capacity.

Internal heating

Field	Default	Units	Status	Meaning
useInternalHeating	false		live	Uniform internal heating rate internalHeating_H (diffuse_temperature).
internalHeating_H	0.0	W kg⁻¹	live	Internal heating rate.
useRadiogenicHeating	false		live	Radiogenic heating with optional decay.
radioHeatingH0	5e-12	W kg⁻¹	live	Bulk rate at $t=0$.
radioHeatingTau	0.0	yr	live	Half-life; 0 = constant.
radioHeatMantleFactor	1.0		live	Multiplier for background mantle.
radioHeatContFactor	10.0		live	Multiplier in continental material.
radioHeatPileFactor	3.0		live	Multiplier in pile material.
useFourElementRadiogenic	false		live	U, Th, K-40 and Al-26 with their own decay (radiogenic_H, utils/radiogenic.jl); replaces radioHeatingH0/radioHeatingTau.
radio_U_ppb, radio_Th_ppb	21.0, 80.0	ppb	live	Uranium and thorium concentration.
radio_K_ppm	250.0	ppm	live	Potassium concentration.
radio_Al26_kgkg	0.0	kg kg⁻¹	live	Al-26 (early solar system only).

Initial conditions

Field	Default	Units	Status	Meaning
rng_seed	42		live	Seed of the marker and noise generator.
m_wavenumber	4		live	Wavenumber of the initial perturbation. It selects the planform the run settles into; quote it with any benchmark result.
dT_anomaly	20.0	K	live	Amplitude of the harmonic perturbation.
noise_amp	5.0	K	live	Random noise amplitude.
r_anomaly	100e3	m	live	Radial extent of the anomaly.
plume_n_min, plume_n_max	0, 0		unused	No reader.
sigma_r	100e3	m	unused	No reader.

The initial temperature profile and boundary temperatures live in TemperatureConditions (below).

Time stepping

Field	Default	Units	Status	Meaning
cflSafety	0.35		live	Courant safety factor (compute_timestep, utils/compute_timestep.jl). Must lie in (0, 1].
dtMax	2e13	s	live	Largest step.
dtMin	1e6	s	live	Smallest step. If CFL asks for less, the floor is applied and counted (cfl_floor_hits()); a run that hits it has abandoned CFL.
maxSteps	10000		live	Step budget.
maxTimeMyr	0.0	Myr	live	Stop at this model time; 0 = no time cap.
advectionScheme	"RK4"		live	Marker advection: "RK4" or "RK2" (advect_markers!).

Markers

Field	Default	Units	Status	Meaning
markersPerCell	9		live	Initial markers per cell (init_markers).
useAdaptiveRemesh	true		live	Conservative reseeding of under- and over-populated cells.
remeshEvery	5	steps	live	Reseeding interval.
reseedNmin	4		live	Cells below this get new markers.
reseedNmax	36		live	Cells above this are culled.

Adaptive mesh refinement

Field	Default	Units	Status	Meaning
useAMR	false		refused	Dynamic refinement is not implemented. validate! raises an error and points to amrStretchFactor.
amrRefineThreshold, amrCoarsenThreshold	0.75, 0.10		inert	Read only by solvers/amr/indicators.jl, which no entry point includes.
amrMaxLevel, amrMinLevel	3, 0		inert	As above.
amrWeight_gradT, amrWeight_strainII, amrWeight_etaContrast, amrWeight_gradV	1.0, 0.5, 0.3, 0.0		inert	As above.

Driver hooks

Field	Default	Status	Meaning
initialMarkerHook	nothing	live	Called once after init_markers as hook(M, grid, cfg, mat, temp). Use it to paint a compositional initial state.
markerStepHook	nothing	live	Called every step after marker advection, before reseeding, as hook(M, r_prev, grid, cfg, step, time_s, dt).
restartHook	nothing	live	Called once after a checkpoint is restored, as hook(M, Tg, grid, cfg, time_myr), for runs that branch from another run's state.

Output

Field	Default	Units	Status	Meaning
outputDir	"output/default"		live	Run directory: log, diagnostics.csv, figures.
bigOutputDir	ENV["MILEVA_BIGOUT"] or ""		live	Where VTK frames and checkpoints go. Empty = outputDir. Resolved by big_output_dir(cfg) (utils/geometry.jl), which falls back to outputDir with a warning if the volume is unreachable.
diagCSV	true		live	Write diagnostics.csv (velocities, mobility, lid thickness, mean T, heat flows, Nusselt numbers, melt ledger, energy budget).
diagEvery	0	steps	live	CSV interval; 0 = logEvery.
logEvery	50	steps	live	Log line interval.
saveVTK	true		live	Write VTK frames (save_vtu).
vtkEvery	500	steps	live	VTK interval.
vtkMinimal	false		unused	No reader.
saveFigures	false		live (2-D)	Six-panel figures (plot_panels, visualization/plot_panels.jl).
framesEvery	500	steps	live	Figure interval.
showFigures	false		unused	No reader.
showTiming	true		unused	No reader; the timing summary is always printed.
saveCheckpoint	false		live	Write JLD2 checkpoints.
checkpointEvery	2000	steps	live	Checkpoint interval.
resumeFrom	""		live	Checkpoint path to resume from; empty = fresh run.
saveMarkers	false		unused	No reader.

TemperatureConditions

Field	Default	Units	Meaning
T_cmb	4000.0	K	CMB temperature (Dirichlet value at the inner face).
T_surf	300.0	K	Surface temperature (Dirichlet value at the outer face).
initial_profile	"adiabatic"		"adiabatic", "conductive" or "uniform". With both TBL thicknesses 0 the adiabatic branch in init_markers falls back to a linear ramp, which is the benchmark initial condition.
tbl_top_thickness, tbl_bottom_thickness	0.0, 0.0	m	Pre-seeded thermal boundary layers of the adiabatic profile.
useLayeredHeating, H_upper, H_lower	false, 5e-12, 1e-12	W kg⁻¹	Depth-layered radiogenic rates.
cmb_cooling, cmb_cooling_rate	false, 100.0	K Gyr⁻¹	Prescribed secular CMB cooling (with cmbHeatfluxMode = "fixed_T").
T_ref	3000.0	K	Floor on an evolving T_cmb.
T_cmb_patches	empty		Gaussian hot or cold patches added to T_cmb (cmb_temperature).

MaterialProperties (summary)

config/MaterialProperties.jl groups the material constants that the viscosity, buoyancy and phase routines read. validate!(mat) computes derived radii. The groups are:

Group	Fields	Used by
Layer radii	R_cmb, R_Dprime, R_lower, R_trans, R_upper, R_asth, R_surf	get_layer, compute_viscosity. Set R_cmb/R_surf to match cfg.R_in/cfg.R_out.
Layer reference viscosities	Dprime_eta0, lower_eta0, trans_eta0, upper_eta0, asth_eta0, lith_eta0, E_act_scale, V_act_scale, T_ref	compute_viscosity (simple path). E_act_scale = 0 makes it isoviscous per layer.
Diffusion creep	diff_eta0, diff_m, diff_d_ref, diff_E_upper, diff_V_upper, diff_E_lower, diff_V_lower	composite_rheology
Dislocation creep	disl_eta0, disl_n, disl_eII_ref, disl_E_upper, disl_V_upper, disl_E_lower, disl_V_lower	composite_rheology
GBS activation	gbs_E_upper, gbs_E_lower	composite_rheology
Yield	maxYieldStress, cohesion_upper, cohesion_lower, friction_upper, friction_lower	composite_rheology (the yield law that runs)
Grain size	grain_k_g, grain_p, grain_E_g, grain_d_min, grain_d_max, grain_C_W, grain_A_pz, grain_q_pz	evolve_grain_size!
Phase transitions	gamma_410, gamma_520, gamma_660, drho_410, drho_520, drho_660, clap_depth_410, clap_depth_520, clap_depth_660, clap_T_ref_410, clap_T_ref_520, clap_T_ref_660, clap_T_ref_ppv, ppv_drho, clap_width_Pa, clap_width_cells	compute_buoyancy, melt_phase. clap_width_Pa = 0 sizes the transition to clap_width_cells radial cells.
Continents	useContinents, continent_theta, continent_thickness, continent_eta_factor, continent_cohesion, continent_friction	marker initialisation (rafts from the surface to continent_thickness); composite_rheology (viscosity factor; Drucker-Prager cohesion and friction of continent material, mixed with the layer values linearly in the continent fraction, NaN keeps the layer values); density through cfg.crustDensityAnomaly
Piles	tcpile_drho, tcpile_eta_factor, tcpile_C_min	compute_buoyancy, viscosity
Composition factors	harzburgite_eta_factor, basalt_eta_factor	viscosity (depletion stiffening, basalt softening)
Published flow laws	flowlaw_diffusion, flowlaw_dislocation, flowlaw_peierls, flowlaw_water, flowlaw_grain_size, flowlaw_crust_diffusion, flowlaw_crust_dislocation, flowlaw_crust_water	composite_rheology; keys and parameters on the Flow laws page. :none keeps the reference-viscosity laws.
Not read (inert)	ppv_eta_factor, R_410, R_520, R_660	No source file outside config/MaterialProperties.jl reads them.

Checking a configuration

validate!(cfg) (config/SimConfig.jl) asserts R_out > R_in, Nr ≥ 4, Nt ≥ 8, a known geometry, cmbHeatfluxMode, compressibilityMode and solverType, 0 < dtMin < dtMax and 0 < cflSafety ≤ 1, and refuses the three flags that cannot do what they promise (useAMR, useDruckerPrager, useMPI in 2-D). disp_summary(cfg) prints a one-screen summary. tests/test_config_matrix.jl exercises every geometry, mode and advertised flag; tests/test_config_timeloop.jl runs each advertised configuration through the real time loop.